Last Season Was A Long Time Ago: Pitcher Half-Season to Half-Season Correlations

Think about where you were in September of last year. It feels like a long time ago, doesn’t it? Perhaps you’ve changed since then, just slightly. Maybe you were sitting at a baseball game. I remember sitting in the shade at Camden Yards because the summer’s sun was still beating down and, by that point, I had received my share of sunburns. The memory seems distant as the last blast of polar wind still cuts through March and my light, dog-walking, jacket. Wherever you were and whatever you were doing in September of last year likely seems like a distant memory. Now consider a pitcher like Blake Snell.

We know where the big lefty was. He was closing out a 2023 second-half ERA that ended at a league-best 1.54. I’m sure you remember that. He went on to win the Cy Young award, he had a bizarre combination of walks per nine innings (5.49) and strikeouts per nine innings (11.20), and he left runners on base (LOB%) at a very high rate of 89.5%. Many people considered it a high-wire act flirting with disaster. Regardless, Snell finished the year with a 1.19 WHIP and a 2.25 ERA and was the talk of the offseason as he and his management floated around in free agency negotiation soup. He finally signed with the San Francisco Giants weeks after pitchers had reported for spring training. Now, Snell will ramp up and join the Giants rotation with September 2023 feeling like a long time ago. With six months of off-season now behind us, we are all wondering how Snell and his fellow second-half performers will start the 2024 season.

As we enter the season I have two questions. First, how many of these pitchers will accumulate another 50 or more innings in the first half of the 2024 season? None of Kyle Bradish, Gerrit Cole, Kodai Senga, or Brandon Woodruff will be throwing many, if any, pitches in the first half of the 2024 season due to injury. Does that mean we should consider performance and workload in the second half of a season a risk when drafting before the next season? Second, does the previous season’s second-half performance correlate with the next season’s first-half performance? Javier Assad pitched wonderfully for the Cubs in the 2023 second half. How will he perform in the first half of 2024?

NOTE: Whenever I ask these questions I find that Jeff Zimmerman has already answered most of them. My analysis is a more specific and current analysis. Here is a link to Zimmerman’s 2014 hitter and pitcher analysis of half-season stat correlations. 

To begin answering these questions I collected data on pitchers who threw at least 50 innings in the second half of the 2018, 2021, and 2023 seasons. I then merged in the next season’s first-half data to make comparisons. Here, for example, are the pitchers whose ERAs fell within a quarter of a point of each other between 2022’s second half and 2023’s first half:

So, there! It can be done. These 10 pitchers accumulated 50 innings or more in the 2022 second half and the 2023 first half and they repeated their performance! But these 10 pitchers represent only eight percent of the 122 that collected at least 50 innings in 2022’s second half. How many of those 122 pitchers threw another 50+ innings in the first half of 2023? That number is 43, or 35%. Here is a table that summarizes IP repeatability from year to year:

Is it rare to repeat innings counts in these small half-to-half samples? Rare isn’t the right word. But, the sub-75% repeat percentage keeps the prospect of using end-of-season inning counts to predict the start of the following season workload a challenge. These IP marks, from half-season to half-season, don’t correlate very strongly among those who reach at least 50 in each:

The pitchers in the ERA repeaters table above, pitchers like George Kirby, Clayton Kershaw, and Marcus Stroman repeated and repeated with great numbers. Pitchers like Nick Pivetta and Miles Mikolas may have hurt your fantasy team ratios. Just because a pitcher repeated from the back end of one season to the front end of another didn’t make them great pitchers. They were just consistent, picking up where they left off. But in the grand scheme of things, how often does that happen? Below is a scatter plot showing second-half ERA and the next season’s first-half ERA. The data includes 2018-2019, 2021-2022, and 2022-2023 for pitchers repeating the 50 IP threshold.

The correlation between second-half ERA and the next season’s first-half ERA is too small to work with, but so is this sample. In reality, there is too much noise in this analysis. The sample is small because I’m too lazy to manually download .csv files. The 2020 season screwed up our ability to generate consistent samples. There’s inherited bias given that teams fighting for a playoff spot behave differently from teams that have lost their chance earlier in the season. Roles change. ERA is flawed. You get the gist. However, if we only focused on the first question of the repeatability of workload, we still would have learned a lot. It’s tough for a major league pitcher to go all out in the second half of an MLB season, cool down, ramp back up, and then do it again in the next season. Last year’s second-half performance should be diluted in your pre-season analysis. Not doing any pre-season analysis? Consider it good bar talk ammo for your Guardians fan friend who continues to soak in the excellent second-half Tanner Bibee displayed in 2023. Ask them how much they would be willing to bet Bibee does it again in 2024’s first half.